Study of logical transaction based data synchronization scheme for pervasive computing

At the center of the data management for pervasive computing is the personalization (i.e. sub-setting) and synchronization of the data for large amount of devices. In this paper we use parameterized database queries to do data sub-setting for our server centric system. We then introduce the concept of logical transaction for synchronization of the data subsets. Server side logical transactions are computed using the state changes of the data subsets. They are used to refresh client side data on devices. Client side logical transactions are tentative, modifiable before they are checked-in on server. In case of conflicts, client application can modify the logical transaction or let server force refresh the client data to resolve the conflicts. This ensures quick data convergence. The logical transaction based synchronization scheme contrasts with traditional physical transaction based replication schemes, which do not apply to subsets of data and cannot scale due to divergence. Test results of an implementation of this scheme show promise of real-world applications.